Problems with bootstrapped zero inflated negative binomial regression that includes a factor variable

Hello,

I'm trying to replicate the ZINB regression results that appear on page 539 of Long & Freese's "Regression Models for Categorical Dependent Variables Using Stata" (3rd edition) using their dataset couart4.dta and changing the specification of the VCE. The original command looks like this:

. zinb art i.female i.married kid5 phd mentor, inf (i.female i.married kid5 phd mentor) nolog

I got very similar results after running the same regression using the option vce(robust):

. zinb art i.female i.married kid5 phd mentor, inf (i.female i.married kid5 phd mentor) vce (robust) nolog

However, when I use the vce(bootstrap) or vce(jackknife) options I get the following message:

. zinb art i.female i.married kid5 phd mentor, inf (i.female i.married kid5 phd mentor) vce (bootstrap) nolog
depvar may not be a factor variable
r(198);

. zinb art i.female i.married kid5 phd mentor, inf (i.female i.married kid5 phd mentor) vce (jackknife) nolog
depvar may not be a factor variable
r(198);

The regression runs OK when I delete the i. prefix from the variables in the inflate part of the model, i.e.:

. zinb art i.female i.married kid5 phd mentor, inf (female married kid5 phd mentor) vce (bootstrap) nolog


... but I need to obtain marginal effects and apparently deleting the i. prefix will lead to wrong results (same book, pages 163-4).

I suspect I'm missing some obvious mathematical/logical point but can't figure out what it is. Any help would be appreciated. I'm using Stata13.

G Vargas